System Logs fill up with i915 errors (1GB per 20 min.)

[BenjaminOtto]

Binary package hint: linux-image-2.6.17-10-generic

On my Lenevo T60 Laptop using Edgy i just noticed, that the system logs fill up very quickly:

ben@riemann:/var/log <0>$ ls -alh /var/log/ | grep M
-rw-r----- 1 root adm 2,3M 2006-10-01 03:13 dpkg.log.1
-rw-r----- 1 root adm 586M 2006-10-10 08:51 kern.log
-rw-r----- 1 root adm 586M 2006-10-10 08:51 messages
-rw-r----- 1 root adm 586M 2006-10-10 08:51 syslog

ben@riemann:/var/log <0>$ ls -alh /var/log/ | grep M
-rw-r----- 1 root adm 2,3M 2006-10-01 03:13 dpkg.log.1
-rw-r----- 1 root adm 594M 2006-10-10 08:52 kern.log
-rw-r----- 1 root adm 594M 2006-10-10 08:52 messages
-rw-r----- 1 root adm 594M 2006-10-10 08:52 syslog

with errors like these:
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(351)
[17182423.712000] i915_emit_cmds(354)
[17182423.712000] i915_emit_cmds(356)
[17182423.712000] i915_emit_cmds(359)
[17182423.712000] i915_emit_cmds(362)
[17182423.712000] validate_cmd( 2000003(0) ): 1
[17182423.712000] i915_emit_cmds(365)
[17182423.712000] i915_emit_cmds(359)
[17182423.712000] i915_emit_cmds(362)
[17182423.712000] validate_cmd( 0(0) ): 1
[17182423.712000] i915_emit_cmds(365)
[17182423.712000] i915_emit_cmds(351)
[17182423.712000] i915_emit_cmds(354)
[17182423.712000] i915_emit_cmds(356)
[17182423.712000] i915_emit_cmds(359)
[17182423.712000] i915_emit_cmds(362)
[17182423.712000] validate_cmd( 54f00006(2) ): 8
[17182423.712000] i915_emit_cmds(365)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(369)
[17182423.712000] i915_emit_cmds(374)
[17182423.712000] i915_emit_cmds(351)
[17182423.712000] i915_emit_cmds(354)
[17182423.716000] i915_emit_cmds(356)
[17182423.716000] i915_emit_cmds(359)
[17182423.716000] i915_emit_cmds(362)
[17182423.716000] validate_cmd( 2000003(0) ): 1
[17182423.716000] i915_emit_cmds(365)
[17182423.716000] i915_emit_cmds(359)
[17182423.716000] i915_emit_cmds(362)
[17182423.716000] validate_cmd( 0(0) ): 1
[17182423.716000] i915_emit_cmds(365)

[BenjaminOtto (benjaminotto)]

Maybe this is the same Bug as #64548 ? I dont know.

[BenjaminOtto (benjaminotto)]

Sometimes Lines like this are in /var/log/messages

i915_emit_cmds(374)
Oct 10 08:10:39 localhost kernel: [171801225555555(((((((((((()))222222222222222222222222222222222222222222222222222222222222
22222222222222222222
Oct 10 08:10:39 localhost kernel:
Oct 10 08:10:39 localhost last message repeated 2 times
Oct 10 08:10:39 localhost kernel: 7ttt
Oct 10 08:10:39 localhost kernel:
Oct 10 08:10:39 localhost last message repeated 8 times
Oct 10 08:10:39 localhost kernel: 0t
Oct 10 08:10:39 localhost kernel:
Oct 10 08:10:39 localhost last message repeated 2 times
Oct 10 08:10:39 localhost kernel: 0ttttttt(999sss 11111l....44444440cccccm 1111111111111594949494949494949494944949
4949211112222
Oct 10 08:10:39 localhost kernel:
Oct 10 08:10:39 localhost last message repeated 8 times
Oct 10 08:10:39 localhost kernel: 222222iiiiiiiiiimmmmmmmmmmmm3333
Oct 10 08:10:39 localhost kernel:
Oct 10 08:10:39 localhost last message repeated 2 times
Oct 10 08:10:39 localhost kernel: 222222222.252022222222222222222222222222222<<<<4_____________c]]]]]]00000000000 59494949492
11112222
Oct 10 08:10:39 localhost kernel:
Oct 10 08:10:39 localhost last message repeated 8 times
Oct 10 08:10:39 localhost kernel: 222222iiiiiiiiiiiimmmmmmmmmmmm3333
Oct 10 08:10:39 localhost kernel:
Oct 10 08:10:39 localhost last message repeated 2 times
Oct 10 08:10:39 localhost kernel: 2222222222222222222222222222222222222222224444 5146920mmmmmmmmmmmmmmmmmmmt00002222222>>2222
555555555555552___dimmmmmdiiiv22iiiiiiiiiiv66660088]]]]]]00000000 594920mmmt00000000777777 (((((((((((75555555555i...........
............111l___c0000000000tttttt
Oct 10 08:10:39 localhost kernel:
Oct 10 08:10:39 localhost last message repeated 10 times
Oct 10 08:10:39 localhost kernel: 22222222222266660000______________________________________c]]]]]]]]]]]]0000000000001111)ddd
ddddddddddddddddd[[[[6666666666666______6767676767676767676767676767676767<<<<1111111111111111)33333333333(mmmd
                                                                       11116)))))))))))))0iiiiie000000000000111111111111>>>>1
)))0mmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmt00007777777777777 (((((75555555_emi55
i22222222222211117777 sssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssss 111l.............
4444443ccccccccccm 0000....<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<f_

[Soren Hansen (soren)]

I'm seeing this too. It stops if I disable beryl, though.

[Soren Hansen (soren)]

Oh, I don't have all those weird messages. "Just" the i915_emit_cmds and validate_cmd.

[Soren Hansen (soren)]

I've located the bug. A bunch of debugging stuff has been thrown into drivers/char/drm/i915_dma.c in i915_emit_cmds() and validate_cmd(). It's nowhere to be found in the git repository, so it's probably a mistake by the uploader.

[Soren Hansen (soren)]

Oh, this also totally breaks dri. I used to get around 500 fps in glxgears. Now I've got 10. :-)

[BenjaminOtto (benjaminotto)]

I am using the compiz packages from http://gandalfn.wordpress.com/ (made from the official compiz-tree) so i should in fact be an issue with i945 and compiz/beryl.

[Soren Hansen (soren)]

It's not a compiz nor beryl issue. They just suffer from the issue as well since they rely DRI to work. Anyone trying to run tuxracer, GL based screensavers or anything else requiring DRI will also have problems.

[Ben Collins (ben-collins)]

Got some local debug in there. Uploaded a fixed source.

[BenjaminOtto (benjaminotto)]

How long will it take for a new binary kernel with this fix being released?

[Soren Hansen (soren)]

On Tue, Oct 10, 2006 at 08:39:29PM -0000, BenjaminOtto wrote:
> How long will it take for a new binary kernel with this fix being
> released?

It shouldn't be more than a few hours at most.

Cheers.

[BenjaminOtto (benjaminotto)]

Ahh, got the new kernel and it seems to work fine again now. Thanks a lot.